Background
The incidence of dementia is increasing rapidly as the population ages [
1,
2]. Alzheimer disease (AD) and dementia with Lewy bodies (DLB) are the two most common subtypes of neurodegenerative dementia [
3]. While AD and DLB have overlapping clinical features, AD is characterized by impairment in memory and learning, executive function and aphasia, whilst the core clinical features of DLB include visual hallucinations, cognitive fluctuations, rapid eye movement (REM) sleep behavior disorder (RBD) and parkinsonism [
3,
4]. The clinical presentation of DLB can vary widely between patients in terms of disease onset, disease progression, treatment response and adverse side effects.
Protein misfolding, accumulation and aggregation are the primary pathological hallmarks of both DLB and AD [
4]. In DLB, abnormal alpha-synuclein (α-syn) accumulates and aggregates in neurons, leading to the formation of Lewy bodies (LBs) and Lewy neurites (LNs), which are typically found in the brainstem, limbic and neocortical areas, spinal cord and peripheral nervous system [
5]. AD is characterized by the deposition of extracellular amyloid beta (Aβ) plaques and intracellular phosphorylated tau (p-tau) tangles [
6,
7]. Approximately 50–80% of DLB cases exhibit AD co-pathology, while circa 50% of AD cases show α-syn pathology in the amygdala and other limbic regions, as observed in clinicopathological studies [
8‐
10]. On top of that, within Lewy body disorders (LBD), the amygdala is suggested to be prone to even initiate the development of α-syn pathology [
11] and function as an incubator for coexisting pathologies such as Aβ and p-tau [
12]. Accumulating evidence from disease models reports interactions between α-syn, p-tau and Aβ pathology, which together form a destructive feed-forward loop towards more severe and faster neurodegeneration [
13‐
16]. Compared to separate pathology, the combination of α-syn, p-tau, and Aβ semi-quantitative pathology scores best predict cognitive decline, as measured by the MMSE [
17]. In addition, DLB patients with AD co-pathology show a highly aggressive disease course with rapid cognitive decline compared to pure DLB cases [
4,
18‐
20]. More importantly, aggregation of α-syn, p-tau, and Aβ pathology is associated with the activation of innate and adaptive immune responses in the elderly [
21].
It is well established that, in AD, the innate immune system is activated in early stages of the disease, and reactive microglia and astrocytes surrounding Aβ plaques are frequently described [
22‐
25]. Regional differences in glial activation have been reported in PD and DLB cases compared to controls, with some studies suggesting a higher activated microglial load, as measured by HLA-DR or CD68, in the amygdala [
26], hippocampus [
27,
28], transentorhinal cortex [
27,
29], and temporal cortex [
27,
30], whereas other studies did not describe an increase in the hippocampus [
31] or neocortex [
32]. Postmortem studies using Iba1 have reported the absence of microglial activity in DLB cases compared to AD cases or controls in the hippocampus [
31] and cerebral cortex [
30,
32]. Only two post-mortem studies of astroglial activation have reported an increase in GFAP expression in the temporal cortex [
33] and pulvinar [
34] in DLB cases compared to controls. The inflammatory response in DLB cases with AD co-pathology has not yet been elucidated, although clarifying these pathomechanisms serves several purposes, such as developing biomarkers to improve early diagnosis, predict disease progression, or to discover targets for early disease-modifying drugs [
35]. Targeting glial cells has been proposed to be beneficial in early-stage neurodegenerative diseases, as neurodegeneration and neuronal cell death are irreversible and are thought to result from chronic neuroinflammation [
36‐
38].
Recent cerebrospinal fluid (CSF) biomarker studies suggest that inflammation in DLB is related to AD co-pathology and is therefore less pronounced in pure DLB cases [
39,
40]. For example, stratification of a DLB group with coexisting AD pathology in the CSF demonstrated higher levels of the glial marker YKL-40 [
39], suggesting that the increase is related to AD neurodegeneration. In addition, a post-mortem study described that the number of activated microglial cells in PD was not associated with the number of LBs [
41]. Moreover, one postmortem study in DLB reported that biochemical measures of CD200 and ICAM-1 correlated with AD plaque density, and found direct co-localization of microglia with AD plaques rather than with LBs [
41,
42]. Interestingly, α-syn accumulation is also observed in activated astrocytes in PD and DLB which may suggest a direct link between α-syn accumulation and activation of the innate immune system [
11,
43‐
45]. In addition, PET imaging studies with 11 C-PK11195, a marker of microglial activation [
46], showed significantly higher binding in early DLB cases when compared to those with an advanced disease stage [
47,
48]. Likewise, 11 C-PK11195 binding was increased in early AD, suggesting that microglial activation is an early event in both DLB and AD [
49,
50]. However, not all studies distinguish between pure DLB and mixed DLB + AD cases, i.e. cases with mixed DLB and AD pathology, when investigating immune responses, leading to conflicting results.
Several genetic studies highlight the relevance of the immune system in AD and PD. For example, single nucleotide polymorphisms (SNPs) in the HLA genetic loci have been associated with a protective effect for both diseases, and are primarily driven by polymorphisms present in most
HLA-DRB1*04 subtypes [
51‐
53]. In addition, genetic risk factor
APOEε4, known to increase the presence of AD co-pathology [
54], was found to be associated with a greater expression of microglial markers CD68 and HLA-DR and reduced expression of Iba1 [
55]. Contrarily, mutations in the glucocerebrosidase (GBA) gene are associated with pathologically pure forms of DLB without AD co-pathology [
56]. For that reason, we believe it is important to investigate the frequency of
HLA-DRB1*04,
APOEε4 and
GBA1 and their effect on pathology and glial load in our cohort.
Our study aims to determine whether there is a different pattern of neuroinflammation in DLB cases with AD co-pathology compared to pure DLB and pure AD cases. In addition, we aim to investigate whether microglial and astroglial activation is associated with α-syn pathology or with AD co-pathology in mixed DLB + AD. Next, we will investigate whether microglial and astroglial activation is higher in regions with a higher burden of α-syn and concomitant p-tau and Aβ pathology. Lastly, we genotyped three common genetic risk genes in DLB and AD, i.e.
GBA1 mutation and frequency of
APOEε4 and
HLA-DRB1*04 alleles, to investigate their effect on pathology burden and inflammatory response in our cohort [
52,
56,
57]. To answer these questions, immunostaining, quantitative image analysis, confocal microscopy and genotyping were performed in limbic and cortical brain regions in a cohort of mixed DLB + AD, pure DLB, pure AD and controls.
Discussion
Microglial reactivity in DLB associated strongly with Aβ and p-tau loads, while no association with astrocytic response was observed. Morphologically, amoeboid and reactive microglia were abundant in mixed DLB + AD and pure AD cases, while pure DLB and control cases mainly revealed homeostatic microglia and small astrocytes with thin processes. Reactive microglial load was higher in mixed DLB + AD compared to pure DLB cases, but DLB phenotypes, i.e. pure and mixed DLB cases, did not differ in astrocytic load. Finally, the highest microglial activity was found in the CA2 and amygdala, where concomitant α-syn and p-tau pathology also showed the highest loads.
Although we observed higher α-syn loads both in cortical regions and the amygdala in mixed DLB + AD cases than in pure DLB cases, which is consistent with previous studies [
78‐
80], significance was only reached in the amygdala in our cohort. These findings suggest that the mixed DLB + AD cases are in a more advanced pathological stage of disease, according to the pattern of α-syn pathology distribution by Braak et al. [
81]. However, DLB phenotypes were scored as Braak α-syn stage 5 or 6 and were not significantly different. This may be explained by the fact that Braak staging does not require evaluation of lesional density and assesses the presence of LBs and LNs [
81], whereas we assessed a quantitative measure of total α-syn pathology. We showed that α-syn load was associated with reactive CD68-positive microglial load only in mixed DLB + AD cases. While several cell-culture and animal studies demonstrated a direct link between α-syn and microglial activation [
37], few post-mortem studies were able to show this association in the cingulate cortex of PD [
27], and in the transentorhinal cortex of pure and mixed DLB cases [
29]. Furthermore, double-staining of HLA-DR microglia and α-syn only showed an association with 20% of the LBs in the cingulate cortex of PD [
27], and it has been suggested that LBs alone are not sufficient to activate microglia [
37]. Moreover, α-syn load strongly associated with astrocytic response in the amygdala in our cohort, a region where astrocytic α-syn was previously reported to be predominant [
11]. In addition, the presence of astrocytic α-syn in the amygdala and other brain regions has recently been demonstrated by studying various post-translational modifications (PTMs) of α-syn [
43]. These observations suggest that reactive microglia and astrocytes might partly respond to the increase of α-syn pathology in mixed DLB + AD cases, or contribute to a more aggressive synucleinopathy.
As previously described and confirmed by our study, early affected regions in AD exhibited similar Aβ- and p-tau- pathological loads in pure AD and mixed DLB + AD cases [
4], while regions affected in more advanced disease stages [
6], such as limbic regions for p-tau and cortical regions for Aβ pathology, were more heavily affected in pure AD cases [
4,
82]. The presence of p-tau did show a positive association with reactive microglial load for pure AD, pure DLB and mixed DLB + AD. Likewise, a recent study using 3D confocal microscopy demonstrated that p-tau was associated with microglial morphological features in the hippocampus of DLB and AD donors [
21]. Interestingly, we observed an overall negative association between Aβ pathology (including diffuse plaques) and microglial load in pure AD and mixed DLB + AD cases. Previous studies have indicated that microglia are able to engulf extracellular fibrillary Aβ plaques [
83]; therefore, a negative association was expected. However, we observed that the association between Aβ pathology and reactive microglial load varied in different brain regions within mixed DLB + AD cases. A negative association was observed in the cornu ammonis, a region with a low Aβ load, while the association was reversed in cortical regions. Early microglial activation in AD has been shown to play a role in the phagocytosis of fibrillary Aβ plaques and precedes a pro-inflammatory harmful state of microglia in later disease stages [
84]. These findings suggest that reactive microglia are able to successfully engulf and clear fibrillary Aβ plaques in regions that are affected only in later disease stages [
84], such as the cornu ammonis, and that an overload of unengaged fibrillary Aβ plaques is observed in regions where the accumulation of Aβ plaques begins in early disease stages. The co-localization of microglia and Aβ plaques has been previously studied by Boon et al., in which the direct co-localization of clustered microglia in classic-cored plaques in AD was demonstrated via immunofluorescence multi-labeling staining [
59]. Similarly, via detailed CLSM, we observed that Aβ plaques co-localize directly with reactive amoeboid microglia in limbic and cortical regions [Fig.
5.].
Previous studies likewise reported that Iba1-positive microglia were highly present in control cases and represented mainly homeostatic microglia, and when double-labeling IHC was performed, mainly reactive swollen amoeboid Iba1-positive microglia were CD68-positive [
85]. In addition, previous studies have described the appearance of Iba1-positive microglia in the healthy aging brain as unchanged [
30,
32] or even more prevalent than in DLB or AD [
31,
63,
86]. However, the significance of the increase in total microglial load with age in the CNS is still unclear [
87]. Presumably, this is independent of the presence of pathology as it did not associate with pathological load. Therefore, a neuroprotective role, rather than a role in promoting disease progression, of Iba1-positive microglia could be hypothesized.
Conflicting results on activated CD68-positive or HLA-DR-positive microglia in DLB phenotypes have been reported previously, in which some studies found an increase in the amygdala, hippocampus, transentorhinal cortex and TC compared to controls [
26,
27,
29,
30], while others did not report an increase in the hippocampus or neocortex [
31,
32]. Besides, detailed transcriptomic analysis of post-mortem brain tissue in DLB failed to show microglial activation [
34,
88]. We only observed a higher reactive microglial load in the CA2 in pure DLB cases compared to controls in our cohort. However, it is important to note that only one study distinguished pure DLB from mixed DLB + AD cases, reporting a higher load in mixed DLB than in pure DLB cases [
29]. We found a similar pattern in our cohort, where mixed DLB + AD cases had a higher reactive microglial load than pure DLB cases in the amygdala and PHG. In addition, we confirmed the higher amoeboid and reactive microglial density in mixed DLB when compared to pure DLB cases. The contradicting results on microglial activity in DLB phenotypes are likely related to the fact that pure and mixed DLB cases were not separated in most cohorts. Moreover, results on microglial upregulation in DLB were based on studies with a small sample size, including only 5 DLB cases in each cohort [
27,
29,
30], and different techniques for quantification were used between studies. Additionally, we found strong associations between microglial activation and AD-pathology in mixed DLB + AD, pure AD and in pure DLB cases, whereas an association between microglial activation and α-syn pathology did not exist in pure DLB cases. In addition, overall microglial activation was the highest in regions with the highest burden of concomitant α-syn and p-tau pathology in our cohort, i.e. the CA2 and amygdala.
We confirmed significant upregulation of astrocytes in pure AD cases [
89,
90]. Two studies in DLB reported an increased astrocytic response in the TC [
33] and pulvinar [
34] compared to controls, which we could not confirm in our results. However, α-syn accumulation in activated astrocytes has been reported in several studies [
11,
37,
91,
92], suggesting a direct link between α-syn accumulation and activation of the innate immune system. Minor astrocytic differences between mixed and pure DLB cases in our cohort suggest astrocytic response to be mostly related to α-syn and not to AD co-pathology, and a region-specific association with α-syn in the amygdala suggests regional vulnerability [
11]. Moreover, we did not find a high astroglial load in regions with a high burden of concomitant pathology. It is important to understand that despite being one of the most extensively used astrocytic markers, GFAP only labels the intermediate filament of the cytoskeleton of mature astrocytes, and is therefore not able to stain all astrocytes [
93].
Confirming previous studies [
94], we found a higher Aβ load in
APOEε4 carriers. In addition, microglial upregulation has previously been described in
APOEε4 carriers [
57], which we confirmed in homozygous
APOEε4 carriers. In line with previous results [
56], we demonstrated that pathogenic
GBA1 mutation carriers have less AD pathology and are most common in pure phenotypes of DLB. Subtypes of
HLA-DRB1*04 were previously described to protect against both PD and AD by improving immune clearance of NFTs [
52]. However, we had a total of 11 cases with
HLA-DRB1*04 alleles in our cohort only, and were not able to reproduce these findings.
Overall, our study highlights the important role of microglial activation, specifically in mixed cases. Importantly, diverse clinical trials that either influence microglial activation in early disease stages, suppress pro-inflammatory responses of microglia or modulate microglial phenotypic changes to support anti-inflammatory capacities in AD are currently being investigated [
95]. In addition, reliable biomarkers to measure AD pathology in CSF exist [
96], supporting the opportunity to make an appropriate selection of mixed DLB cases for future clinical trials on immunomodulatory approaches.
Strengths of this study include being one of the first to investigate co-pathology and inflammation in a large number of brain regions in both mixed and pure DLB in a qualitative, semi-quantitative and quantitative manner. Furthermore, multiple neuroinflammatory markers were evaluated, microglia response near plaques using confocal microscopy were studied, and the effects of common genotypes on neuropathology and inflammation were analyzed. However, although we studied a very well-defined cohort, the sample size and clinical information on disease severity were limited. Further, other pathologies, such as TDP-43, and molecular phenotypes of glial cells, to better understand their activation state, were not studied. Future research should focus on spatial transcriptomics to study the molecular phenotype of reactive microglia and astroglia surrounding (co-)pathology. Besides, the quantitative QuPath analyses detected the total %area of immunopositivity, and therefore detected all positive structures. An Artificial Intelligence based approach that would be able to distinguish the different morphological pathological and glial structures would be of great interest for future studies. Finally, studying the level of AD pathology and neuroinflammation via in-vivo biomarkers, e.g., in CSF or plasma, would enhance the investigation of a more comprehensive spectrum of disease stages.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.